Why Software Estimates Are Always Wrong

As a senior engineer, I often found myself estimating big projects in six weeks. This tendency came to a head during my time at Amazon Tickets, our ticket-selling business akin to Ticketmaster. I vividly recall being tasked with estimating the timeline for a complete overhaul of our ordering pipeline.

At that time, our ticket-selling capacity was constrained by external partners, limiting our ability to handle sales for major events. The project involved redesigning our ordering system to integrate directly with Amazon's backend, which we knew would meet our scaling needs.

Leveraging my familiarity with the existing system, I broke down the problem into smaller, manageable components that could be developed in parallel. Given our fully staffed team, lack of competing priorities, and my personal involvement in designing the new system, I estimated we could deliver in six weeks. This timeline would allow ample time for integration testing and instill confidence for an upcoming major ticke…